Persistent Effects of Herbaceous Species.pdf - d-Commons
Persistent Effects of Herbaceous Species on the Infectious Lethality of Soil for Conifer Seedlings David J. Schimpf Department of Biology and Olga Lakela Herbarium, University of Minnesota Duluth Duluth, MN 55812 USA [email protected] Steven C. Garske1 Department of Biology, University of Minnesota, Duluth, Minnesota 55812 USA [email protected] Ronald R. Regal Department of Mathematics and Statistics, University of Minnesota, Duluth, Minnesota 55812 USA [email protected] 1Present address: Great Lakes Indian Fish and Wildlife Commission, Odanah, Wisconsin 54861, USA. Key words: Aegopodium podagraria, apparent competition, damping‐off, epidemiology, indirect effects, plant disease, Weibull model Abstract Seeds of the coniferous trees Abies balsamea, Picea mariana, and Pinus strobus were sown in the laboratory in two soils taken from ground‐layer patches differing in species composition, one of which was dominated by Aegopodium podagraria (goutweed). This permitted inference whether herbaceous species may affect the favorableness of the soil for establishment of these trees. Weibull distributions were fitted to the time course of aggregate seedling emergence and post‐emergence mortality, enabling seedling lifespan to be inferred without monitoring of each individual. A higher percentage of Abies seeds developed into emerged seedlings in the goutweed soil, likely because of less pre‐ emergence mortality incited by pathogens. Picea and Pinus emergence percentages were similar in both soils. Most emerged seedlings died within weeks, with symptoms of diseases incited by soil‐ or seed‐borne fungi. Although the timing of seedling emergence did not differ between soils, seedlings died more quickly on the goutweed soil, largely because of a faster development of post‐emergence damping‐off. Total post‐emergence mortality of Picea and Pinus was greater on the goutweed soil. The
2
relative frequencies of the several symptoms exhibited by dying seedlings varied between the two soils, suggesting that the ground‐layer species differentially affected the microbial community’s composition or interactions with the seedlings. Symptom frequencies differed among tree species. Local spatial variation in herbaceous species composition appeared to produce patchy infectious lethality of soil for tree seedlings, an indirect effect that was observed after the herbaceous plants had been removed. Introduction The recruitment of forest tree populations from seed may be inhibited by herbaceous species or other ground‐layer plants (Daniel et al. 1979, George and Bazzaz 2003, Maguire and Forman 1983). For tree species tolerant of overstory shade, this may limit advance regeneration. Suppression of tree seedling establishment by the ground layer could come about through competition for limiting resources, release of directly inhibitory allelochemicals, apparent competition ‐ the ground layer’s stimulation of other species that are the tree seedlings’ natural enemies (Holt and Lawton 1994), or the ground layer’s inhibition of other species that benefit the tree seedlings. Because apparent competition and inhibition of beneficial species are mediated by third parties, each represents an indirect interaction (Wooton 1994) between neighboring plant species. An extreme degree of local dominance by the forb Aegopodium podagraria L. (goutweed) has been observed in one place where it was introduced into forest vegetation beyond its native Eurasian range. At the site of a former settlement in the southern boreal forest, Minnesota, USA, a 0.23‐ha continuous cover of goutweed under an evergreen coniferous overstory was found to have low vascular plant richness and diversity, in comparison to that in an adjoining native forest ground layer (Garske 2000). Seedling and sapling densities of shade‐tolerant coniferous trees were greatly depressed within the goutweed patch, which had existed for several decades before the ‐1 data of Garske (2000) were obtained (Ahlgren and Ahlgren 1984). There were 3340 ha ‐1 of these juvenile trees amidst goutweed, vs. 18,200 ha amidst the native ground layer (Garske 2000). All of the juvenile trees in the sample quadrats were the shade‐tolerant ‐1 Abies balsamea (L.) Miller. If we use 2000 trees ha as a representative regional value for a fully stocked overstory of Abies and Picea with stem diameter greater than 10 cm (Bakuzis and Hansen 1965), this allows for little further mortality of juvenile trees in the goutweed patch before the future overstory may become discontinuous. Ahlgren and Ahlgren (1984) proposed that allelopathy accounted for the extreme dominance by goutweed on this site. We would add that goutweed appears to be an overbearing competitor for limiting light. We observed that goutweed produces a dense layer of closely abutting leaflets that is held higher than the leaves of conifer seedlings and most native forest herbs at the Minnesota boreal site, allowing goutweed to pre‐empt photons
3
where light is scarce. A goutweed patch forms by lateral extension of rhizomes. Details on the structure and dynamics of goutweed clones are given by Gatsuk et al. (1980). Our study attempted to discover whether herbaceous species, especially goutweed, differentially suppress tree recruitment through processes beyond resource competition. Because early seedling development is an especially vulnerable portion of the life cycle of many species, we focused on possible effects of the ground layer on seedling establishment through its influence on soil characteristics. The opportunity for a seedling to acquire soil nutrients generally has little influence on its early success at establishment (Harper 1977), so any ground‐layer effects on nutrient availability may be assumed to have low direct importance for this phase of the seed‐plant life cycle. Any observed influence on establishment by ground‐layer conditioned soil could then be attributed to allelopathy or to indirect interactions. We tested the hypotheses that contrasting ground‐layer composition influenced soil properties so as to affect the percentage of seedling emergence, the time course of seedling emergence, the percentage of post‐emergence seedling survival, and the time course of post‐emergence seedling survival. Methods Soil source Because soil could not be brought promptly to the laboratory from the remote boreal site described above, soil was obtained from a human‐modified northern deciduous forest site in Duluth, Minnesota, USA (46°48´ N 92°8´ W 400 m elevation), in the first week of June, 1994, after herbs had grown to nearly full size. One soil collection was made within a dense 220 m2 patch of goutweed; the other soil was collected 1 ‐ 2 m beyond the edge of the goutweed patch, with both collection areas on the same contour. Only goutweed grew in the ground layer where “goutweed soil” was collected. The “other soil” was collected near the non‐native species Phalaris arundinacea L. and Valeriana officinalis L., where the ground‐layer canopy was less continuous and extended over a much greater vertical range. Both ground‐layer patches had a discontinuous overstory of Populus balsamifera L. about 8 m tall. Soil was removed by excavating surface blocks about 15 cm square and 7 ‐ 10 cm deep, trimming off the outer 1 cm of soil from the cut edges of the block with a dull knife to exclude possible sap from wounded roots or rhizomes, then gently crumbling the remaining block. This was repeated until the total soil volume was about 0.05 m3 from each type of ground layer. Evident stones, roots, rhizomes, and invertebrates were removed manually, and the remaining soil mixed by hand. Analysis of soil texture (hydrometer method) found the goutweed soil to be 70% sand, 9% silt, and 21% clay; the other soil was 65% sand, 11% silt, and 24% clay (USDA particle sizes). Soil pH (1:1 water), measured with a pH meter (SA720, Thermo Orion, Beverly, MA, USA), was 6.2 for the goutweed soil and 6.3 for the other soil.
4
Seed source Seeds of the conifers Abies balsamea, Picea mariana (Miller) BSP, and Pinus strobus L. were obtained from the Minnesota Department of Natural Resources. Each of these species is highly dependent on seed for recruitment (Burns and Honkala 1990) and was common in the canopy at the boreal field site. The supplier reported the viabilities as 94%, 95%, and 91%, respectively. Each seed lot had been collected from a native population in Minnesota within 32´ latitude and 65 m altitude of the site from which the soil was taken. Seeds were stratified in moist paper towels at 5 °C for 60 days just before planting. Germination experiment Immediately after it was mixed, each soil was distributed into nine plastic flats (25 × 52 × 6 cm) having drainage holes, giving a soil depth of 3.5 cm. Both soils had a crumb structure, with aggregates up to a few mm in diameter. Seeds were sown immediately. Each flat received 48 seeds of each species in an 18 × 8 grid, with every seed being 3 cm from its nearest neighbors. We alternated the species at consecutive points in the planting grid. Seeds were covered with 3 ‐ 6 mm of soil. We covered each flat with a transparent plastic dome that fit loosely enough to allow slight air exchange. Equal amounts of deionized water were added to each flat as a fine mist periodically throughout the experiment, which kept the soil moist but not wet enough to result in a visible sheen of liquid. Flats were placed on table tops with the two soils alternating in spatial sequence. The room was illuminated for 12 h per day by cool‐white fluorescent ceiling lamps, and its temperature fluctuated within 21 ‐ 26 °C. Counts of live and dead emerged seedlings were made daily for 45 days, except for day 44. Seedlings were not monitored individually. The covers were removed from the flats while observations were made. Each dead seedling was assigned to one category of symptoms based on Hartley et al. (1918): (1) classic damping‐off (mechanical failure of the hypocotyl near the soil surface), (2) moldy shoot (visible mycelium enveloping the epicotyl and cotyledons or attached seed coat), or (3) top wilt (seedling erect and shriveled or discolored, but not visibly moldy). These symptoms are associated with infections by fungi (sensu lato). The date of death was the first daylight period when (1) the seedling had fallen over, (2) the mycelium had reached the shoot apex and base of the cotyledons, or (3) the cotyledons were shriveled or discolored to their base, respectively. Each dead seedling was left in place, and a 1‐cm thread was laid on the soil next to it on the day of death to enable differentiation of new fatalities from old ones. At the end of the experiment, instantaneous photosynthetic photon flux density (PPFD) was measured. A point quantum sensor (LI‐190SA and LI‐1000, LICOR, Lincoln, NE, USA) was held horizontally 15 cm in from each end of each flat and 4 cm above the soil surface, and the two readings per flat were averaged. The sample mean ‐2 ‐1 and 95% confidence interval (CI) PPFD was 6 ± 0.6 μmol m s for each kind of soil. For each flat the soil was then mixed, and a sample with a moist mass of about 250 g had its
5
water content determined gravimetrically (Θm) at the end of the experiment. The sample mean and 95% CI Θm was 0.33 ± 0.03 for the goutweed‐soil flats and 0.35 ± 0.02 for the flats with the other soil. Data analysis The time‐course of a population’s germination or seedling emergence can be represented closely by a Weibull distribution (Brown and Mayer 1988). Weibull distributions can also be used to analyze disease progress in plants (Campbell 1998). We built Weibull models of seedling emergence and post‐emergence mortality with SAS (SAS Institute 2004) (Appendix). The fit of the model function to the mean of the daily observations was tested with an approximation of the Komolgorov ‐ Smirnov test that used a critical value (0.05 level) of 1.36/n0.5. From the model outputs we estimated the final mean proportion emerged, mean time to emergence of the 50th percentile of the sown seeds, mean time alive after emergence for the 25th percentile of emerged seedlings that died, and final proportion of emerged seedlings that survived. Standard errors and tests of significance were computed with 400 bootstrap runs. In accord with potential statistical dependence of plants in the same flat, bootstraps were obtained by randomly sampling flats with replacement, rather than individual plants. Frequencies of mortality symptoms were compared between soils within species by multi‐response permutation procedure (MRPP), and frequencies of symptoms were compared between species within soils by blocked MRPP, both using PC‐ORD (McCune and Mefford 1999). In both types of analysis, counts were relativized per flat, and Euclidean distance and n/Σ(n) weighting were used. For the blocked MRPP, median alignment was performed. Post‐emergence days until death for each symptom was compared between soils within species with a t‐test, using Statistix (Analytical Software 2003). Results Observations of cumulative seedling emergence and post‐emergence mortality are summarized in Figure 1. The emergence trajectories were very similar between soils within species, being most gradual for Abies and most abrupt for Picea. The fit of the model functions (Table 1) to the means of the emergence observations was within the critical value of 0.065 for all species, soils, and days, with two exceptions: for Pinus on the other soil (soil 2), the difference between the model and the observed mean was 0.075 on day 12 and 0.089 on day 13. These were days when cumulative emergence was increasing especially rapidly, a situation anticipated by Brown and Mayer (1988). From model estimates of total emergence (pe, Table 1), a greater fraction of Abies seeds developed into emerged seedlings on the goutweed soil than on the other soil (bootstrap P = 0/400). The fraction of Picea seeds that developed into emerged seedlings was not significantly different between the two soils, and this was also true for Pinus. Picea seedlings emerged earlier than the other two species. Pinus 50th‐percentile
6
Abies soil 1
50
Emerged Dead
40
30
20
20
10
10
0
0
10
20
30
40
50
Picea soil 1
50
Cumulative seedlings
40
30
0
0
40
30
30
20
20
10
10
0
0
10
20
30
40
50
Pinus soil 1
50
30
30
20
20
10
10
0
0
10
20
30
10
40
50
30
40
50
20
30
40
50
40
50
Pinus soil 2
50 40
20
Picea soil 2
0
40
0
10
50
40
0
Abies soil 2
50
0
10
20
30
Day
Figure 1. Cumulative seedling emergence (open circles) and cumulative mortality of emerged seedlings (solid circles) for Abies balsamea, Picea mariana, and Pinus strobus sown on two different forest soils in the laboratory. Soil 1 was from a patch of goutweed, soil 2 from other herbaceous species. There were 48 seeds of each tree species in each of nine flats per soil, sown on day 0. Bars represent ±1 standard deviation.
7
emergence occurred about 4 days later than that of Picea, and that of Abies occurred about 9 days later than that of Picea. Within species, the time of 50th‐percentile emergence was very similar on both soils (Table 1). The appearance of the post‐emergence mortality trajectories (Figure 1) differs between soils within species, and among species within soils. Mortality began sooner on the goutweed soil. It began soonest and accumulated most rapidly for Picea, latest and most slowly for Abies. The fit of the models (Table 1) to the means of post‐ emergence mortality observations was within the critical value in every case. Total post‐emergence survival (ps) of Picea was only one‐third as high on goutweed soil as on the other soil, a statistically significant difference (bootstrap P = 0.015) (Table 1). Total post‐emergence survival of Pinus seedlings was four to six times as high as that of Picea. Total survival for Pinus was about one‐half as high on goutweed soil as on the other soil; this seemingly large difference was not statistically significant, apparently because the delayed mortality on the other soil censored the data sufficiently to expand the standard error of the estimate. Similarly, the even later occurrence of mortality for Abies did not permit meaningful estimates of total post‐emergence survival for that species (Table 1). Early post‐emergence mortality occurred sooner on goutweed soil. The estimated time after emergence until the first 25% of total mortality was complete was shorter on goutweed soil by about 5 days (Abies) or 9 days (Picea and Pinus) (Table 1). Each of these between‐soil differences was statistically significant (bootstrap P = 0/400). All newly dead seedlings had symptoms of infections by fungi (sensu lato). Classic damping‐off was the most frequent symptom for all three species on both soils, and was substantially more frequent on goutweed soil than on the other soil for both Picea and Pinus (Table 2). The between‐soil trend for Abies was the same, but the effect size was smaller and the difference not statistically significant. On the other soil the lower frequencies of classic damping‐off were balanced mainly by higher frequencies of moldy shoot for Pinus and of top wilt for Picea. For each of the three species‐pair comparisons of frequency distributions, the effect size was less on the goutweed soil, although the difference between Abies and Picea was not statistically significant on either soil. The differences in frequency distributions were statistically significant on each soil for Abies vs. Pinus and for Picea vs. Pinus. For each of the three species, mortality from classic damping‐off tended to occur earlier on the goutweed soil than on the other soil (Table 3). The statistical significance of this is attributable in part to greater test power associated with much larger numbers of observations of this symptom, yet the between‐soil absolute difference in mean days between sowing and death was also larger for classic damping‐off than it was for the other two categories of symptoms for each of the species. Mean date of observed mortality from classic damping‐off occurred earlier on goutweed soil by 3.3 days for Abies, 9.1 days for Picea, and 10.4 days for Pinus. Mean date of mortality of Picea from
Table 1. Estimated parameters (see Appendix) for seedling emergence and mortality. Soil 1 was from a patch of goutweed, soil 2 from other herbaceous species. Numbers in parentheses are standard error of the mean. Percentile emerged is days after sowing, percentile dead is days after emergence. For statistical significance, see text. Parameter Species and Soil______________________________ Abies soil 1 Abies soil 2 Picea soil 1 Picea soil 2 Pinus soil 1 Pinus soil 2 z1 7.33 8.41 6.91 6.77 3.45 6.74 2.00 1.85 2.12 2.48 4.48 2.58 c1 k1 0.0708 0.0762 0.262 0.258 0.0835 0.117 0.000 2.06 1.35 3.08 6.83 0.000 z2 2.68 4.14 1.72 2.63 1.53 3.16 c2 k2 0.0258 0.0269 0.0937 0.0548 0.0923 0.0366 Total emergence (pe) 0.85 (0.014) 0.77 (0.018) 0.91 (0.018) 0.92 (0.016) 0.96 (0.011) 0.93 (0.018) 50th percentile emerged (days) 19.1 (0.35) 19.2 (1.02) 10.1 (0.10) 10.1 (0.038) 14.5 (0.10) 14.1 (0.26) 25th percentile dead (days) 24.3 (0.99) 29.5 (0.95) 6.26 (0.32) 14.9 (0.35) 12.4 (0.44) 21.7 (1.22) n.a. n.a. 0.03 (0.012) 0.09 (0.019) 0.18 (0.034) 0.35 (0.13) Total post‐emergence survival (ps) ________________________________________________________________________________________________________
9
Table 2. Proportion of dead seedlings assigned to mutually exclusive categories of symptoms. Soil 1 was from a patch of goutweed, soil 2 from other herbaceous species. Means are followed by standard deviations in parentheses. A can be regarded as effect size in MRPP, P is probability of an equal or smaller delta. Species and soil Category Classic damping‐off Moldy shoot Top wilt Abies soil 1 0.893 (0.141) 0.011 (0.033) 0.096 (0.117) Abies soil 2 0.842 (0.142) 0.087 (0.112) 0.070 (0.086) Picea soil 1 0.947 (0.038) 0.022 (0.020) 0.030 (0.024) Picea soil 2 0.788 (0.087) 0.030 (0.024) 0.182 (0.078) Pinus soil 1 0.886 (0.059) 0.090 (0.049) 0.026 (0.024) 0.044 (0.050) Pinus soil 2 0.664 (0.072) 0.293 (0.085) Abies, soil 1 vs. 2 A = 0.009, P = 0.294 Picea, soil 1 vs. 2 A = 0.384, P
2
relative frequencies of the several symptoms exhibited by dying seedlings varied between the two soils, suggesting that the ground‐layer species differentially affected the microbial community’s composition or interactions with the seedlings. Symptom frequencies differed among tree species. Local spatial variation in herbaceous species composition appeared to produce patchy infectious lethality of soil for tree seedlings, an indirect effect that was observed after the herbaceous plants had been removed. Introduction The recruitment of forest tree populations from seed may be inhibited by herbaceous species or other ground‐layer plants (Daniel et al. 1979, George and Bazzaz 2003, Maguire and Forman 1983). For tree species tolerant of overstory shade, this may limit advance regeneration. Suppression of tree seedling establishment by the ground layer could come about through competition for limiting resources, release of directly inhibitory allelochemicals, apparent competition ‐ the ground layer’s stimulation of other species that are the tree seedlings’ natural enemies (Holt and Lawton 1994), or the ground layer’s inhibition of other species that benefit the tree seedlings. Because apparent competition and inhibition of beneficial species are mediated by third parties, each represents an indirect interaction (Wooton 1994) between neighboring plant species. An extreme degree of local dominance by the forb Aegopodium podagraria L. (goutweed) has been observed in one place where it was introduced into forest vegetation beyond its native Eurasian range. At the site of a former settlement in the southern boreal forest, Minnesota, USA, a 0.23‐ha continuous cover of goutweed under an evergreen coniferous overstory was found to have low vascular plant richness and diversity, in comparison to that in an adjoining native forest ground layer (Garske 2000). Seedling and sapling densities of shade‐tolerant coniferous trees were greatly depressed within the goutweed patch, which had existed for several decades before the ‐1 data of Garske (2000) were obtained (Ahlgren and Ahlgren 1984). There were 3340 ha ‐1 of these juvenile trees amidst goutweed, vs. 18,200 ha amidst the native ground layer (Garske 2000). All of the juvenile trees in the sample quadrats were the shade‐tolerant ‐1 Abies balsamea (L.) Miller. If we use 2000 trees ha as a representative regional value for a fully stocked overstory of Abies and Picea with stem diameter greater than 10 cm (Bakuzis and Hansen 1965), this allows for little further mortality of juvenile trees in the goutweed patch before the future overstory may become discontinuous. Ahlgren and Ahlgren (1984) proposed that allelopathy accounted for the extreme dominance by goutweed on this site. We would add that goutweed appears to be an overbearing competitor for limiting light. We observed that goutweed produces a dense layer of closely abutting leaflets that is held higher than the leaves of conifer seedlings and most native forest herbs at the Minnesota boreal site, allowing goutweed to pre‐empt photons
3
where light is scarce. A goutweed patch forms by lateral extension of rhizomes. Details on the structure and dynamics of goutweed clones are given by Gatsuk et al. (1980). Our study attempted to discover whether herbaceous species, especially goutweed, differentially suppress tree recruitment through processes beyond resource competition. Because early seedling development is an especially vulnerable portion of the life cycle of many species, we focused on possible effects of the ground layer on seedling establishment through its influence on soil characteristics. The opportunity for a seedling to acquire soil nutrients generally has little influence on its early success at establishment (Harper 1977), so any ground‐layer effects on nutrient availability may be assumed to have low direct importance for this phase of the seed‐plant life cycle. Any observed influence on establishment by ground‐layer conditioned soil could then be attributed to allelopathy or to indirect interactions. We tested the hypotheses that contrasting ground‐layer composition influenced soil properties so as to affect the percentage of seedling emergence, the time course of seedling emergence, the percentage of post‐emergence seedling survival, and the time course of post‐emergence seedling survival. Methods Soil source Because soil could not be brought promptly to the laboratory from the remote boreal site described above, soil was obtained from a human‐modified northern deciduous forest site in Duluth, Minnesota, USA (46°48´ N 92°8´ W 400 m elevation), in the first week of June, 1994, after herbs had grown to nearly full size. One soil collection was made within a dense 220 m2 patch of goutweed; the other soil was collected 1 ‐ 2 m beyond the edge of the goutweed patch, with both collection areas on the same contour. Only goutweed grew in the ground layer where “goutweed soil” was collected. The “other soil” was collected near the non‐native species Phalaris arundinacea L. and Valeriana officinalis L., where the ground‐layer canopy was less continuous and extended over a much greater vertical range. Both ground‐layer patches had a discontinuous overstory of Populus balsamifera L. about 8 m tall. Soil was removed by excavating surface blocks about 15 cm square and 7 ‐ 10 cm deep, trimming off the outer 1 cm of soil from the cut edges of the block with a dull knife to exclude possible sap from wounded roots or rhizomes, then gently crumbling the remaining block. This was repeated until the total soil volume was about 0.05 m3 from each type of ground layer. Evident stones, roots, rhizomes, and invertebrates were removed manually, and the remaining soil mixed by hand. Analysis of soil texture (hydrometer method) found the goutweed soil to be 70% sand, 9% silt, and 21% clay; the other soil was 65% sand, 11% silt, and 24% clay (USDA particle sizes). Soil pH (1:1 water), measured with a pH meter (SA720, Thermo Orion, Beverly, MA, USA), was 6.2 for the goutweed soil and 6.3 for the other soil.
4
Seed source Seeds of the conifers Abies balsamea, Picea mariana (Miller) BSP, and Pinus strobus L. were obtained from the Minnesota Department of Natural Resources. Each of these species is highly dependent on seed for recruitment (Burns and Honkala 1990) and was common in the canopy at the boreal field site. The supplier reported the viabilities as 94%, 95%, and 91%, respectively. Each seed lot had been collected from a native population in Minnesota within 32´ latitude and 65 m altitude of the site from which the soil was taken. Seeds were stratified in moist paper towels at 5 °C for 60 days just before planting. Germination experiment Immediately after it was mixed, each soil was distributed into nine plastic flats (25 × 52 × 6 cm) having drainage holes, giving a soil depth of 3.5 cm. Both soils had a crumb structure, with aggregates up to a few mm in diameter. Seeds were sown immediately. Each flat received 48 seeds of each species in an 18 × 8 grid, with every seed being 3 cm from its nearest neighbors. We alternated the species at consecutive points in the planting grid. Seeds were covered with 3 ‐ 6 mm of soil. We covered each flat with a transparent plastic dome that fit loosely enough to allow slight air exchange. Equal amounts of deionized water were added to each flat as a fine mist periodically throughout the experiment, which kept the soil moist but not wet enough to result in a visible sheen of liquid. Flats were placed on table tops with the two soils alternating in spatial sequence. The room was illuminated for 12 h per day by cool‐white fluorescent ceiling lamps, and its temperature fluctuated within 21 ‐ 26 °C. Counts of live and dead emerged seedlings were made daily for 45 days, except for day 44. Seedlings were not monitored individually. The covers were removed from the flats while observations were made. Each dead seedling was assigned to one category of symptoms based on Hartley et al. (1918): (1) classic damping‐off (mechanical failure of the hypocotyl near the soil surface), (2) moldy shoot (visible mycelium enveloping the epicotyl and cotyledons or attached seed coat), or (3) top wilt (seedling erect and shriveled or discolored, but not visibly moldy). These symptoms are associated with infections by fungi (sensu lato). The date of death was the first daylight period when (1) the seedling had fallen over, (2) the mycelium had reached the shoot apex and base of the cotyledons, or (3) the cotyledons were shriveled or discolored to their base, respectively. Each dead seedling was left in place, and a 1‐cm thread was laid on the soil next to it on the day of death to enable differentiation of new fatalities from old ones. At the end of the experiment, instantaneous photosynthetic photon flux density (PPFD) was measured. A point quantum sensor (LI‐190SA and LI‐1000, LICOR, Lincoln, NE, USA) was held horizontally 15 cm in from each end of each flat and 4 cm above the soil surface, and the two readings per flat were averaged. The sample mean ‐2 ‐1 and 95% confidence interval (CI) PPFD was 6 ± 0.6 μmol m s for each kind of soil. For each flat the soil was then mixed, and a sample with a moist mass of about 250 g had its
5
water content determined gravimetrically (Θm) at the end of the experiment. The sample mean and 95% CI Θm was 0.33 ± 0.03 for the goutweed‐soil flats and 0.35 ± 0.02 for the flats with the other soil. Data analysis The time‐course of a population’s germination or seedling emergence can be represented closely by a Weibull distribution (Brown and Mayer 1988). Weibull distributions can also be used to analyze disease progress in plants (Campbell 1998). We built Weibull models of seedling emergence and post‐emergence mortality with SAS (SAS Institute 2004) (Appendix). The fit of the model function to the mean of the daily observations was tested with an approximation of the Komolgorov ‐ Smirnov test that used a critical value (0.05 level) of 1.36/n0.5. From the model outputs we estimated the final mean proportion emerged, mean time to emergence of the 50th percentile of the sown seeds, mean time alive after emergence for the 25th percentile of emerged seedlings that died, and final proportion of emerged seedlings that survived. Standard errors and tests of significance were computed with 400 bootstrap runs. In accord with potential statistical dependence of plants in the same flat, bootstraps were obtained by randomly sampling flats with replacement, rather than individual plants. Frequencies of mortality symptoms were compared between soils within species by multi‐response permutation procedure (MRPP), and frequencies of symptoms were compared between species within soils by blocked MRPP, both using PC‐ORD (McCune and Mefford 1999). In both types of analysis, counts were relativized per flat, and Euclidean distance and n/Σ(n) weighting were used. For the blocked MRPP, median alignment was performed. Post‐emergence days until death for each symptom was compared between soils within species with a t‐test, using Statistix (Analytical Software 2003). Results Observations of cumulative seedling emergence and post‐emergence mortality are summarized in Figure 1. The emergence trajectories were very similar between soils within species, being most gradual for Abies and most abrupt for Picea. The fit of the model functions (Table 1) to the means of the emergence observations was within the critical value of 0.065 for all species, soils, and days, with two exceptions: for Pinus on the other soil (soil 2), the difference between the model and the observed mean was 0.075 on day 12 and 0.089 on day 13. These were days when cumulative emergence was increasing especially rapidly, a situation anticipated by Brown and Mayer (1988). From model estimates of total emergence (pe, Table 1), a greater fraction of Abies seeds developed into emerged seedlings on the goutweed soil than on the other soil (bootstrap P = 0/400). The fraction of Picea seeds that developed into emerged seedlings was not significantly different between the two soils, and this was also true for Pinus. Picea seedlings emerged earlier than the other two species. Pinus 50th‐percentile
6
Abies soil 1
50
Emerged Dead
40
30
20
20
10
10
0
0
10
20
30
40
50
Picea soil 1
50
Cumulative seedlings
40
30
0
0
40
30
30
20
20
10
10
0
0
10
20
30
40
50
Pinus soil 1
50
30
30
20
20
10
10
0
0
10
20
30
10
40
50
30
40
50
20
30
40
50
40
50
Pinus soil 2
50 40
20
Picea soil 2
0
40
0
10
50
40
0
Abies soil 2
50
0
10
20
30
Day
Figure 1. Cumulative seedling emergence (open circles) and cumulative mortality of emerged seedlings (solid circles) for Abies balsamea, Picea mariana, and Pinus strobus sown on two different forest soils in the laboratory. Soil 1 was from a patch of goutweed, soil 2 from other herbaceous species. There were 48 seeds of each tree species in each of nine flats per soil, sown on day 0. Bars represent ±1 standard deviation.
7
emergence occurred about 4 days later than that of Picea, and that of Abies occurred about 9 days later than that of Picea. Within species, the time of 50th‐percentile emergence was very similar on both soils (Table 1). The appearance of the post‐emergence mortality trajectories (Figure 1) differs between soils within species, and among species within soils. Mortality began sooner on the goutweed soil. It began soonest and accumulated most rapidly for Picea, latest and most slowly for Abies. The fit of the models (Table 1) to the means of post‐ emergence mortality observations was within the critical value in every case. Total post‐emergence survival (ps) of Picea was only one‐third as high on goutweed soil as on the other soil, a statistically significant difference (bootstrap P = 0.015) (Table 1). Total post‐emergence survival of Pinus seedlings was four to six times as high as that of Picea. Total survival for Pinus was about one‐half as high on goutweed soil as on the other soil; this seemingly large difference was not statistically significant, apparently because the delayed mortality on the other soil censored the data sufficiently to expand the standard error of the estimate. Similarly, the even later occurrence of mortality for Abies did not permit meaningful estimates of total post‐emergence survival for that species (Table 1). Early post‐emergence mortality occurred sooner on goutweed soil. The estimated time after emergence until the first 25% of total mortality was complete was shorter on goutweed soil by about 5 days (Abies) or 9 days (Picea and Pinus) (Table 1). Each of these between‐soil differences was statistically significant (bootstrap P = 0/400). All newly dead seedlings had symptoms of infections by fungi (sensu lato). Classic damping‐off was the most frequent symptom for all three species on both soils, and was substantially more frequent on goutweed soil than on the other soil for both Picea and Pinus (Table 2). The between‐soil trend for Abies was the same, but the effect size was smaller and the difference not statistically significant. On the other soil the lower frequencies of classic damping‐off were balanced mainly by higher frequencies of moldy shoot for Pinus and of top wilt for Picea. For each of the three species‐pair comparisons of frequency distributions, the effect size was less on the goutweed soil, although the difference between Abies and Picea was not statistically significant on either soil. The differences in frequency distributions were statistically significant on each soil for Abies vs. Pinus and for Picea vs. Pinus. For each of the three species, mortality from classic damping‐off tended to occur earlier on the goutweed soil than on the other soil (Table 3). The statistical significance of this is attributable in part to greater test power associated with much larger numbers of observations of this symptom, yet the between‐soil absolute difference in mean days between sowing and death was also larger for classic damping‐off than it was for the other two categories of symptoms for each of the species. Mean date of observed mortality from classic damping‐off occurred earlier on goutweed soil by 3.3 days for Abies, 9.1 days for Picea, and 10.4 days for Pinus. Mean date of mortality of Picea from
Table 1. Estimated parameters (see Appendix) for seedling emergence and mortality. Soil 1 was from a patch of goutweed, soil 2 from other herbaceous species. Numbers in parentheses are standard error of the mean. Percentile emerged is days after sowing, percentile dead is days after emergence. For statistical significance, see text. Parameter Species and Soil______________________________ Abies soil 1 Abies soil 2 Picea soil 1 Picea soil 2 Pinus soil 1 Pinus soil 2 z1 7.33 8.41 6.91 6.77 3.45 6.74 2.00 1.85 2.12 2.48 4.48 2.58 c1 k1 0.0708 0.0762 0.262 0.258 0.0835 0.117 0.000 2.06 1.35 3.08 6.83 0.000 z2 2.68 4.14 1.72 2.63 1.53 3.16 c2 k2 0.0258 0.0269 0.0937 0.0548 0.0923 0.0366 Total emergence (pe) 0.85 (0.014) 0.77 (0.018) 0.91 (0.018) 0.92 (0.016) 0.96 (0.011) 0.93 (0.018) 50th percentile emerged (days) 19.1 (0.35) 19.2 (1.02) 10.1 (0.10) 10.1 (0.038) 14.5 (0.10) 14.1 (0.26) 25th percentile dead (days) 24.3 (0.99) 29.5 (0.95) 6.26 (0.32) 14.9 (0.35) 12.4 (0.44) 21.7 (1.22) n.a. n.a. 0.03 (0.012) 0.09 (0.019) 0.18 (0.034) 0.35 (0.13) Total post‐emergence survival (ps) ________________________________________________________________________________________________________
9
Table 2. Proportion of dead seedlings assigned to mutually exclusive categories of symptoms. Soil 1 was from a patch of goutweed, soil 2 from other herbaceous species. Means are followed by standard deviations in parentheses. A can be regarded as effect size in MRPP, P is probability of an equal or smaller delta. Species and soil Category Classic damping‐off Moldy shoot Top wilt Abies soil 1 0.893 (0.141) 0.011 (0.033) 0.096 (0.117) Abies soil 2 0.842 (0.142) 0.087 (0.112) 0.070 (0.086) Picea soil 1 0.947 (0.038) 0.022 (0.020) 0.030 (0.024) Picea soil 2 0.788 (0.087) 0.030 (0.024) 0.182 (0.078) Pinus soil 1 0.886 (0.059) 0.090 (0.049) 0.026 (0.024) 0.044 (0.050) Pinus soil 2 0.664 (0.072) 0.293 (0.085) Abies, soil 1 vs. 2 A = 0.009, P = 0.294 Picea, soil 1 vs. 2 A = 0.384, P
